when is a housing bubble not a housing bubble? -...
TRANSCRIPT
When is a Housing Bubble Not a Housing Bubble?
Margaret Hwang SmithDepartment of Economics
Pomona CollegeClaremont CA 91711
Gary SmithDepartment of Economics
Pomona CollegeClaremont CA 91711
909-607-3135fax: 909-621-8576
Chris ThompsonDepartment of Economics
Pomona CollegeClaremont CA 91711
Corresponding author: Gary Smith
This research was supported by the John Randolph Haynes and Dora Haynes Foundation
When is a Housing Bubble Not a Housing Bubble?
Abstract
Discussions of housing bubbles generally rely on indirect barometers such as rapidly increasing
prices, unrealistic expectations of future price increases, and rising ratios of housing price indexes
to household income and to rent indexes. These indirect measures cannot answer the key question
of whether housing prices are justified by the anticipated cash flow. We show how to estimate
the fundamental value of a house and use unique rent and price data for matched homes in several
southern California cities to illustrate this approach. Our evidence indicates that the bubble is
not, in fact, a bubble in that, under a variety of plausible assumptions, buying a house at current
market prices still appears to be an attractive long-term investment.
When is a Housing Bubble Not a Housing Bubble?
U. S. housing prices have risen 50% in the past five years, and more than 100% in some hot
markets. Many knowledgeable observers believe that we are in the midst of a speculative bubble
in residential real estate prices, particularly on the coasts, that rivals the dot-com bubble in the
1990s and that will have a similarly unhappy conclusion.
In December 2004, UCLA Anderson Forecast’s Economic Outlook described the California
housing market as a bubble, repeating their warnings made in previous years. Yale economist
Robert Shiller has issued similar housing-bubble alarms for several years and, in June 2005,
warned that, “The [housing] market is in the throes of a bubble of unprecedented proportions
that probably will end ugly.” Shiller suggests that real housing prices might fall by 50% over the
next decade. In August 2005, Princeton’s Paul Krugman argued that there was definitely a
housing bubble on the coasts and that, indeed, the air had already begun leaking out of the bubble.
The evidence that has been cited in support of a housing bubble has been suggestive, but
indirect, in that it does not address the key question of whether housing prices are justified by
the anticipated cash flow. We show how to estimate the fundamental value of a house and use a
unique set of rent and price data for matched homes in several southern California cities to
illustrate this approach. Our evidence indicates that, even though prices have risen rapidly and
many buyers have unrealistic expectations of continuing price increases, the bubble is not, in fact,
a bubble in that, under a variety of plausible assumptions, buying a house at current market
prices still appears to be an attractive long-term investment.
What is a Bubble?
Charles Kindleberger (1987) defined a bubble this way:
A bubble may be defined loosely as a sharp rise in price of an asset or a range of assets in a
1
continuous process, with the initial rise generating expectations of further rises and
attracting new buyers—generally speculators interested in profits from trading rather than
in its use or earning capacity. The rise is then followed by a reversal of expectations and a
sharp decline in price, often resulting in severe financial crisis—in short, the bubble bursts.
Researchers often focus on specific aspects of this general concept: rapidly rising prices (Baker
2002), unrealistic expectations of future price increases (Case and Shiller 2003); the departure of
prices from fundamental value (Garber 2000), or a large drop in prices after the bubble pops
(Siegel 2003)
We define a bubble as a situation where the market prices of a certain class of assets (such as
stocks or real estate) rise far above the present value of the anticipated cash flow from the asset
(Kindleberger’s use or earning capacity). This definition suggests many of the features noted
above: prices rising rapidly, a speculative focus on future price increases rather than the asset’s
cash flow, and an eventual drop in market prices. However, these features are only suggestive.
Market prices can rise rapidly if fundamental values are increasing rapidly. Large future price
increases can be realistically anticipated during inflationary periods. Market prices can drop (for
example, in a financial crisis) even when there has been no bubble. What truly defines a bubble is
that the market prices are not justified by the asset’s cash flow.
One appealing way to answer this question is, as with dividend-discount models of stock
prices, assume that the investment is for keeps—that the buyer never sells and is therefore
unconcerned about future prices. Few people literally plan to hold stocks or houses forever, but
this assumption allows us to determine whether the cash flow alone is sufficient to justify the
current market price. We will use this approach and we will also look at some finite horizons
with modest assumptions about future prices; for example, annual increases of 0% to 3% over
the next decade.
2
Is the Housing Market Efficient?
True believers in efficient markets might deny that there can ever be a bubble. The market
price is always the correct price and is therefore justified by the expectations market participants
hold. Even Siegel, who believes there can be bubbles, writes that, “We know that the price of any
asset is the present value of all future expected cash flows.” Contrast this with the opening
sentence of John Burr Williams’ classic treatise, The Theory of Investment Value (1938):
“Separate and distinct things not to be confused, as every thoughtful investor knows, are real
worth and market price.” In the stock market, these two arguments can perhaps be reconciled by
a consideration of whether the future expected cash flows investors use to calculate present
values are reasonable. The residential real estate market is fundamentally different in that
homebuyers do not calculate present values.
Case and Shiller (2003) report survey evidence of homeowners’ naive beliefs about the real
estate market. The residential real estate market is populated by amateurs making infrequent
transactions on the basis of limited information and with little or no experience in gauging the
fundamental value of the houses they are buying and selling. It is highly unlikely that residential
real-estate prices are always equal to the present value of the expected cash flow if market
participants almost never attempt to estimate the present value of the expected cash flow.
Instead, the nearly universal yardstick in residential real estate is “comps,” the recent sale
prices of nearby houses with similar characteristics. Comps tell us how much others have paid
for houses recently, but not whether these prices are justified by the cash flow. Is a Britannia the
British Bear Beanie Baby worth $500 because a Princess Beanie Baby sold for $500? Is this
house worth $1 million because a similar house sold for $1 million? The nearly universal use of
comps by buyers, sellers, real estate agents, and appraisers is the very mechanism by which
market prices can wander far from fundamental values. If no one is estimating fundamental value,
3
why should we assume that market prices will equal fundamental values?
In the stock market, professional investors can, in theory, arbitrage and exploit the mistakes
made by noise traders. In the housing market, however, professionals cannot sell houses short
and cannot obtain the tax advantages available for owner-occupied housing by physically
occupying multiple houses. If a myopic focus on comps causes market prices to depart from
fundamentals, there is no effective self-correcting mechanism.
Of course, in an inefficient market, prices can be above or below fundamental value. Housing
bubble enthusiasts implicitly assume that market prices were equal to fundamental values in the
past and that recent increases have pushed prices above fundamental values. Perhaps housing
prices were too low in the past and recent increases have brought market prices more in line with
fundamentals.
Supply and Demand
Researchers have used a variety of proxies to gauge whether real estate prices have departed
from fundamental values. For example, Case and Shiller (2003) look at the ratio of housing prices
to household income, the idea being that housing prices are a bubble waiting to pop if the median
buyer is priced out of the market. But the affordability of a house has nothing to do with its
intrinsic value. Berkshire Hathaway stock currently sells for nearly $100,000 a share. It is not
affordable for most investors, but it may be worth the price!
Aggregate measures of housing prices over time are notoriously imperfect because houses are
so heterogeneous in their characteristics and location; it is difficult to measure depreciation and
remodeling of existing homes; and it is difficult to measure changes in the quality of home
construction over time. McCarthy and Peach (2004 ) show that, between 1977 and 2003, four
popular home price indexes showed price appreciation ranging from 199% (constant quality new
homes) to 337% (median price of sales of existing homes).
4
Even on its own terms, the ratio of housing prices to income doesn’t really measure
affordability. A better measure would be the ratio of mortgage payments to income. Mortgage
rates have fallen dramatically and the ratio of mortgage payments on a constant-quality new
home to median family income has fallen steadily, from 0.35 in 1981 to 0.13 in 2003 (McCarthy
and Peach 2004 ).
Other analysts look at various barometers of supply and demand. A supply and demand
analysis of the housing market inevitably leads to a consideration of rents since high housing
prices might be justified by high rents. Would you be willing to pay a high price for rent-
controlled housing in Santa Monica simply because the demand is high and the supply is fixed?
Not unless you expect rents to rise in the future. Thus, restrictions on building, a limited supply
of urban land, an increasing population, and growing incomes of home buyers, are all factors that
might justify high prices, but only insofar as they justify high rents.
Some economists cite the fact that house prices have risen faster than rents as evidence of a
bubble (Leamer 2002, Hatzius 2004). Krainer and Wei (2004) report that there has been a 30%
real increase in house prices over the past decade and a 10% real increase in rents over this same
period, suggesting that prices are departing from fundamentals.
Housing prices and rents are tied together by the fact that the fundamental value of a house
depends on the anticipated rents, in the same way that the fundamental value of bonds and
stocks depends on the present value of the cash flow from these assets. However, just as bond
and stock prices are not tied rigidly to coupons or dividends, so the fundamental value of a house
is not tied rigidly to rents. Among the many factors that affect the price-rent ratio are interest
rates, risk premiums, growth rates, and tax laws (including property taxes and income tax
brackets). Thus, just as with price-earnings ratios in the stock market, price-rent ratios in the
housing market can rise without signaling a bubble if, for example, interest rates fall or there is an
5
increase in the anticipated rate of growth of rents.
In addition, the dwellings included in price indexes do not match the dwellings in rent indexes,
giving us a comparison of apples to oranges (McCarthy and Peach 2004). The ratio of a house
price index to a rent index can rise because the prices of houses in desirable neighborhoods
increased more than the rents of apartment buildings in less desirable neighborhoods. Or perhaps
the quality of the average house in the price index has increased relative to the quality of the
average property in the rent index. And, in any case, to gauge fundamental value, we need actual
rent and price data, not indexes with arbitrary scales.
Buying Versus Renting
Thus we are inevitably drawn back to the need to estimate a house’s use or earning capacity.
Because shelter can be obtained either by renting or buying, the implicit cash flow from an
owner-occupied house is based on the rent that would otherwise be paid to live in the house.
Buying and renting have sometimes been analyzed as demands for different commodities.
Rosen (1979) wrote that, “In many cases it is difficult (say) to rent a single unit with a large
backyard. Similarly, it may be impractical for a homeowner to contract for the kind of
maintenance services available to a renter.” A decade later, Goodman (1988) observed that, “Until
recently, it was easier to purchase small (large) amounts of housing by renting (owning). As a
result, households with tastes for small (large) units would rent (buy).”
Today, it is still true that rental and sale properties differ, on average, in location and
attributes. But, on the margin, close substitutes are generally available. It is possible to buy small
condominiums and to rent houses with large yards. It is possible to buy or rent small or large
houses. Many households have the option of buying houses in communities that provide services
very similar to those received by most renters.
We consequently view buying and renting as often being viable alternatives. If a household has
6
the opportunity to buy or rent very similar properties (perhaps even the same property), then
the relevant question is whether to pay for these housing services by buying the property or
renting it.
Fundamental Value
Rental savings are the central, but not the only, factor in determining the fundamental value of
an owner-occupied house. We have to look at everything that affects the cash flow, including
transaction costs, the down payment, insurance, maintenance, property taxes, mortgage
payments, tax savings, and the proceeds if the house is sold at some point.
Once we have the projected cash flow, we can value houses the same way we value bonds,
stocks, and other assets—by discounting the cash flow by the household’s required rate of
return. Specifically, we can calculate the net present value (NPV) of the entire cash flow,
including the initial outlay:
†
NPV = X0 +X1
1+ R( )1+
X2
1+ R( )2+
X3
1+ R( )3+K+
Xn
1+ R( )n(1)
X0 is a negative number equal to the downpayment and out-of-pocket closing costs. Xn is the net
amount received when the house is sold and the mortgage balance (if any) is paid off. The
intervening cash flows are the rent you would otherwise have to pay to live in this house minus
the expenses associated with home ownership, plus or minus the value of nonfinancial factors
(such as pride of ownership and a desire for privacy). The rent and other expenses can be
estimated from observed data. The intangibles must be assigned values by the household.
The required return R depends on the rates of return available on other investments. The initial
downpayment ties up funds that could otherwise be invested in bonds, stocks, and other assets;
as the years pass, the net rental savings free up funds that can be invested elsewhere. The
7
required return depends on current interest rates but, because there is considerable uncertainty
about the net cash flow from a house, a homebuyer may use a required return similar to that
applied to stocks and comparably risky investments.
A homebuyer can use the projected cash flow and a required rate of return to determine if a
house’s NPV is positive or negative. If the NPV is positive, the house is indeed worth what it
costs; if the NPV is negative, renting is more financially attractive. We can also calculate the
internal rate of return (IRR) that makes the NPV equal to zero. The cash flow from residential
real estate is generally negative in the early years and positive in later years, with just one sign
change, so that the NPV is a monotonically decreasing function of the required return. If so, the
IRR identifies the breakeven required return for which the investor is indifferent about the
investment, and the NPV is positive for any required return less than the IRR.
Data
To illustrate this approach, we gathered data in the summer of 2004 from the Internet and
newspaper advertisements for matched pairs of single-family homes in the San Gabriel Valley,
which is in the eastern portion of Los Angeles County, south of the San Gabriel Mountains. The
San Gabriel Valley was once predominantly agricultural, but after World War II the farmlands
began their now (almost complete) conversion into suburban communities. Today, nearly two
million people (one fifth of the total population of Los Angeles County) live in the San Gabriel
Valley. The two largest cities are Pomona and Pasadena. Among the low-income, middle-income,
and high-income communities are El Monte, Alhambra, and San Marino.
We only looked at detached single-family homes, and excluded apartments, condominiums,
and houses in gated communities. The Yahoo Maps web site was used to determine the driving
distance between properties. The matched rental and sale properties could differ by no more than
100 square feet in size (or by no more than 250 square feet for houses with more than 2500
8
square feet), no more than 1 bedroom, and no more than half a bath. The driving distance between
properties had to be less than 1 mile. The houses were also matched for style (for example,
ranch) and in identified amenities, such as a pool. No doubt the properties often differed in other
ways (carpet versus wood floors, fireplace or no fireplace), but we can hope that these
differences averaged out over our sample. There were often multiple matches (for example, two
sales to one similar rental), and we used the best overall match in terms of square footage,
distance, and so on. In some cases, there was a perfect match in that a house that had been
offered for rent was sold.
If the rental and sale data were for different months, the sales price was adjusted by 1 percent
a month, which is the figure used by appraisers at this time. For example, if the rental was listed
in July and the matching house sold in June, the sale price was adjusted upward by 1%.
We found 139 observations that matched a sale to a rent that was close by and had similar
characteristics. The average distance between the matched properties is 0.33 miles. Table 1
shows the mean characteristics of the properties and the mean of the absolute value of the
difference between the values of the matched properties.
Analysis
The following assumptions were used to determine the cash flow: 20% downpayment, 30-
year mortgage, 5% mortgage rate, closing costs equal to 1% of the mortgage, insurance equal to
0.1% of the house’s price, maintenance equal to 1% of the price, property taxes equal to 1% of
the price, $2400 for annual utilities, 28% federal income tax rate, 9.3% state income tax rate, 15%
capital gains tax (if the capital gains exceed $500,000), 6% required after-tax return, 8% sales
transaction cost (including brokerage fees, closing costs, and fix-up expenses), 3% annual increase
in housing rents and expenses (the average increase in the CPI in the Los Angeles area over the
past 5 years), and 2% annual increase in property taxes (as limited by Proposition 13).
9
One way to gauge whether market prices can be justified without unrealistic expectations
about future prices is to assume that the buyer never sells, analogous to dividend-discount
models of stock prices that assume the investor “buys for keeps.” If the buyer never sells, then
future prices are irrelevant; all that matters is the cash flow from the asset. We also look at finite
horizons with conservative assumptions about future prices. We calculated the NPV and IRR for
an infinite holding period and also for finite holdings periods of up to 30 years with annual price
increases ranging from 0% to 3%.
For example, we found two adjacent houses on Tulsa Avenue in Claremont. Both are 2530
square-foot ranch houses with four bedrooms, three baths, and a pool. One rented for $2650 a
month; the other sold for $622,960. Although we used monthly data, Table 2 shows an annual
summary of the projected cash flow for selected years, with a 3% annual increase in the sale
price. The homebuyer’s anticipated net cash flow is negative for the first 10 years, but the
homeowner is building up equity in an appreciating asset and the after-tax IRR is positive by the
third year. The NPV is negative until the IRR goes above 6%. As the horizon lengthens, the IRR
rises initially as the transaction cost becomes less important, and then peaks and declines slightly
as the tax-deductible mortgage is paid off. The forever horizon does not depend on an assumed
growth rate for the house’s price; the 7.3% after-tax forever IRR is slightly lower than the 7.9%
value for a 30-year horizon with a 3% annual price increase.
Figure 1 shows the NPVs for 5-, 10-, and 20-year horizons using required returns ranging from
0 to 20%. For low required rates of return, the NPVs increase with the holding period because (a)
the mortgage payments are fixed while the net rental savings grow over time, causing the cash
flow to go from negative to increasingly positive; and (b) the homeowner is building up equity in
an appreciating asset. The IRRs are where the NPV curves cross the horizontal line at NPV = 0.
The household should buy if its after-tax required rate of return is less than the IRR and should
10
rent otherwise. After-tax IRRs of 7-9% certainly appear attractive.
Price-Rent Ratios
Commercial real estate is commonly valued by applying a multiple of 5 or 6 times earnings or
EBITDA. For our Tulsa Avenue properties, the annual rent is $31,800 and EBITDA is $31,800 -
$6,230 - $2,400 - $623 - $6,230 = $16,317. The ratio of the price to the annual rent is
$622,960/($31,800) = 19.6 and the ratio of the price to EBITDA is $622,960/($16,317) = 38.2,
which are very high by commercial real estate standards.
In general, we shouldn’t expect a constant multiple to give a theoretically correct fundamental
value for either commercial or residential real estate, since the value of a property depends on
interest rates, growth rates, tax laws, and even the down payment if the after-tax interest rate
used to discount the cash flow is not equal to the after-tax mortgage rate. Similarly, we should not
expect the price-rent ratio to be constant across properties or across time and we should not
expect the rent-price ratio or the EBITDA-price ratio to be an accurate estimate of a house’s
IRR.
Figure 2 shows a scatter plot for all 139 matched pairs of the annual rent-price ratio and the
after-tax IRR for a 10-year holding period with 3% annual price increases. The least squares line
is IRR = -6.6 + 3.0(rent/price), with an R-square of 0.985. The NPV equation is very nonlinear
and it is remarkable that the relationship between the IRR and the rent-price ratio is so tightly
linear. Notice also that not only is the IRR not equal to the rent-price ratio, but the slope is not
equal to 1. For every 1 percentage point increase in the rent-price ratio, the IRR increases by
approximately 3 percentage points. For example, the IRR is 5.4% if the annual rent-price is 4%
and the IRR is 8.4% if the annual rent-price ratio is 5%. Different assumptions about the various
parameters will yield a somewhat different fitted line. For example, if we increase the growth
rates of rent, price, and expenses to 4%, the fitted line is IRR = -3.5 + 2.8(rent/price), again with
11
an R-square of 0.985.
Sensitivity Analysis
We can gauge the robustness of our results by varying our key assumptions. Table 3 shows
the after-tax IRRs for the Tulsa Avenue house with a variety of assumptions regarding the
growth rate of the house’s price, rent, and expenses. The only buy-for-keeps scenario with an
after-tax IRR below 5% is with rents growing by 1% a year forever. (Remember, price growth
doesn’t matter with an infinite holding period.) The after-tax IRRs for finite holding periods are
reasonably attractive, except for a few scenarios, such as a 10-year horizon with 1% growth in
prices, rents, and expenses or a 10-year horizon with 0% price growth.
For another kind of sensitivity analysis, we can incorporate stochastic changes in rent and
prices into the model and use Monte Carlo simulations to estimate probability distributions for
the NPVs and IRRs. We will illustrate this approach here with seemingly reasonable, but
conservative, assumptions about the stochastic parameters. Because our matched rent/price data
are unique, there are no directly comparable historical data that can be used to estimate the
means, variances, and covariances of rent and prices that we need for our simulations. Instead, we
use Los Angeles-area indexes to give ballpark estimates.
The Bureau of Labor Statistics has a consistent index for owner’s equivalent rent of primary
residence back to 1983; the Department of Housing and Urban Development has repeat-sales
price indexes for this same period. Over the years 1983-2004, the annual growth rates of the Los
Angeles-area rent and price indexes were 3.9% and 6.3% respectively, with a correlation of 0.57
(p-value = 0.01). The standard deviations of the annual percentage changes in Los Angeles-area
rents and price were 2.3% and 8.8% respectively.
We will work with monthly cash flows, but assume that rent increases occur every 12-months.
We simplify our Monte Carlo simulations by assuming that price increases also occur at 12-
12
month intervals. We should not assume that the future will replicate the past and, indeed, our
objective here is to see if the purchase of a house can be justified financially even with
conservative assumptions about future increases in rents and prices. Our illustrative calculations
consequently assume that the annual changes in rent and price are lognormally distributed, each
with a 3% mean, with standard deviations of 2% and 9%, respectively, and with a 0.60
correlation.
Specifically, we assume that rent Y and price P are determined by these stochastic equations:
†
Yt = Yt-1(1+ et )
Pt = Pt-1(1+ nt )
where 100
†
et and 100
†
nt are the annual percentage changes in rent and price, and the natural
logarithms of
†
1+ et and
†
1+ nt are jointly normally distributed with means, standard deviations,
and correlation coefficient that correspond to the means, standard deviations and correlation
coefficient of
†
et and
†
nt . For example, if
†
et has a mean m = 0.03 and standard deviation s = 0.02,
then the log of
†
1+ et has this mean:
†
E ln 1+ et[ ][ ] = 0.5ln(1+ m)4
(1+ m)2 + s2
È
Î Í Í
˘
˚ ˙ ˙
= 0.5ln1.034
1.032 + 0.022
È
Î Í Í
˘
˚ ˙ ˙
= 0.02937
For n independent simulations, each with a probability p of the IRR falling within a
prespecified interval, the simulation standard error for the Monte Carlo estimate of p is
approximately
†
p(1- p)n
13
One million simulations were used with a maximum standard error equal to 0.0005.
The fixed-rate columns in Table 4 shows the median IRRs, median NPVs (using a 6% after-tax
required rate of return), and value-at-risk (VaR) for the NPV using a 95% confidence level over
various horizons. A complete distribution of the IRRs is problematic for many horizons because
there are some scenarios in which rents and prices do not rise sufficiently to ever generate a
positive cash flow—which means there is no IRR. The medians reported in Table 4 treat these
problematic IRRs as negative.
As before, the forever NPVs and IRRs do not depend on future housing prices. Figure 3
shows the complete probability distribution for the NPV with an infinite horizon and a 6% after-
tax required return. There is an estimated 1.70% chance that the NPV will be negative (an IRR
will be below 6%) and and a 0.00% chance that the IRR will be negative.
These calculations assume that the homebuyer chooses a 30-year mortgage with a fixed 5%
mortgage rate. Many homebuyers instead choose variable-rate mortgages, perhaps because the
initial interest rate is less than that on a longer-term fixed rate mortgage. There are, of course, a
plethora of fixed-rate and variable-rate options. We will focus on the cash-flow risk inherent in a
variable rate mortgage by assuming that the initial mortgage rate is 5%, the same as with our 30-
year fixed-rate mortgage, and that the mortgage rate is adjusted every 12 months based on the
average interest rate on 1-year Treasury securities during the most recent month, with a 2
percentage-point cap on the annual change in interest rates and a 10% maximum for the interest
rate. Every time the mortgage rate is changed, the loan is amortized over the remaining years—30
years minus the years already elapsed.
For modeling monthly changes in the Treasury rate we use the discrete version of the well-
known Cox, Ingersoll, and Ross (CIR) mean-reverting model: (1985):
†
Rt - Rt-1 = a(b - Rt-1)Dt + Rt-1se Dt
14
where the time interval is
†
Dt = 1/12, the long run equilibrium interest rate is b = 0.05, the pull-
back factor is a = 0.20, the instantaneous standard deviation is s = 0.02, and the stochastic term e
is normally distributed with mean 0 and standard deviation 1.
The Federal Reserve has monthly interest data on 1-year constant maturity Treasury
securities back to April 1954. Since we assume that the adjustments in the mortgage rate are
based on the monthly average interest rate at 12-month intervals, we use the changes in monthly
average Treasury rates at 12-month intervals for guidance. During the years 1983-2004, the
correlation between the annual percentage-point changes in the 1-year Treasury rate and Los
Angeles-area housing prices was 0.11 (p = .61) and the correlation between annual percentage-
point changes in the 1-year Treasury rate and Los Angeles-area rents was -0.16 (p = 0.47). We
can also look at the historical correlations for more frequent data—quarterly for the price index
and monthly for the rent index. The correlation between quarterly percentage-point changes in
the 1-year Treasury rate and Los Angeles-area prices was 0.12 (p = 0.20) and the correlation
between monthly percentage-point changes in the 1-year Treasury rate and Los Angeles-area
rents was -0.01 (p = 0.89). Our Monte Carlo simulations consequently assume that percentage
changes in 1-year Treasury rates are uncorrelated with rents and prices.
The adjustable-rate column in Table 4 shows the median IRRs for the Tulsa Avenue house
over various horizons for 1 million simulations. As before, the forever IRRs do not depend on
future housing prices. There is an estimated 33.20% chance that the buy-for-keeps NPV will be
negative (an IRR will be below 6%) and and an 0.52% chance that the IRR will be negative.
The adjustable-rate case is generally less favorable than the fixed-rate case because of the
increase cash-flow risk and the asymmetry in interest-rate changes. When the index rate declines,
it cannot go below 0%, and this forms a base from which the index rate can increase. However,
15
index rate increases are unbounded and, while the adjustable mortgage rate is capped at 10%, the
index rate can go much higher and consequently keep the mortgage rate at 10% much longer.
Thus, although the adjustable rate starts at 5% and is as likely to rise as fall, the asymmetry in
the boundaries skews the distribution of the adjustable mortgage rate upward. Even so, for
horizons beyond 10 years, the negative NPVs reflect after-tax IRRs between 5% and 6%, which
are reasonably attractive.
Conclusion
For our 139 matched pairs, the rent-price ratios ranged from 3.2% to 9.6%. To gauge whether
the projected cash flow justifies the price, we focus on buying for keeps. With an infinite horizon
(and no assumptions about future prices), the after-tax IRRs with a fixed 5% 30-year mortgage
ranged from 4.3% to 20.5% with a median value of 6.9%. With a finite horizon, and assuming
that prices increase, on average, by 3% a year, the after-tax IRRs range from 2.8% to 23.1% with
a 10-year horizon and from 4.1% to 20.7% with a 30-year horizon. These conclusions are
relatively robust with respect to a plausible range of assumptions about future rents and (for the
finite-horizon case) future prices. If, however, the homebuyer uses an adjustable-rate mortgage,
the median IRRs are somewhat lower and there is a greater chance that the IRR will turn out to be
disappointing.
Housing prices have increased rapidly in the Los Angeles area in recent years and many
homebuyers have unrealistic expectations about future prices. The relevant question, however, is
not how much prices have increased in the past or how fast people expect them to increase in the
future, but whether, at current prices, a house is still a fundamentally sound investment. Our
answer is yes, even if the owner either buys for keeps (with no assumptions about future house
prices) or makes conservative assumptions about future housing prices.
16
References
Baker, Dean, 2002. The Run-Up in Home Prices: A Bubble, Center for Economic and Policy
Research, Challenge 46 (6), 93-119.
Case, Karl E., and Robert J. Shiller, 2003. Is There a Bubble in the Housing Market? An
Analysis, Brookings Brookings Papers on Economic Activity (Brookings Institution), 2003:2,
299-342.
Cox, John. C., Jonathan E. Ingersoll, Jr., and Stephen. A. Ross, 1985. “A Theory of the Term
Structure of Interest Rates,” Econometrica, 53: 385-407.
Garber, Peter, 2000. Famous First Bubbles: The Fundamentals of Early Manias , Cambridge,
Mass.: MIT Press.
Goodman, Allen C., 1988. An Econometric Model of Housing Price, Permanent Income, Tenure
Choice, and Housing Demand, Journal of Urban Economics, 23: 327-353.
Hatzius, Jan, 2004. Housing and the U. S. Consumer: Mortgaging the Economy’s Future,
Goldman Sachs Global Economics Paper no. 83.
Kindleberger, Charles, 1987. Bubbles, The New Palgrave: A Dictionary of Economics, John
Eatwell, Murray Milgate, and Peter Newman, eds., New York: Stockton Press: 281
Krainer, John, and Chishen Wei, 2004. House Prices and Fundamental Value, FRBSF Economic
Letter, 2004-27.
Krugman, Paul, 2005. That Hissing Sound, The New York Times: August 8.
Leamer, Edward E., 2002. Bubble Trouble? Your Home Has a P/E Ratio Too, UCLA Anderson
Forecast.
McCarthy, Jonathan, and Richard W. Peach, 2004. Are Home Prices the Next “Bubble”?,
FRBNY Economic Policy Review, 10 (3): 1-17.
Rosen, Harvey S., 1979. Housing Decisions and the U. S. Income Tax: An Econometric Analysis,
17
Journal of Public Economics, 11: 1-23.
Shiller, Robert, 2005. The Bubble’s New Home, Barron’s: June 20.
Siegel, Jeremy J., 2003. What Is an Asset Price Bubble? An Operational Definition, European
Financial Management, 9 (1): 11-24.
John Burr Williams, 1938. The Theory of Investment Value, Cambridge, Mass.: Harvard
University Press.
18
Table 1 Mean Values for Rental and Sale Properties
Rental Sale Absolute
House House Difference
rent or price $2189/month $528,236
square feet 1808 1801 44
bedrooms 3.4 3.3 0.1
bathrooms 2.1 2.2 0.1
19
Table 2 Cash Flow for Claremont Houses on Tulsa Avenue, sale price increasing 3% annually
Rent Mortgage Property Tax Other Net Cash Net Sales Mortgage NPV IRR
Year Savings Payments Taxes Savings Expenses Flow Price Balance R = 6% (%)
1 31,800 -32,104 -6,230 10,749 -9,253 -5,037 590,317 -491,015 -40,920 -30.7
2 32,754 -32,104 -6,354 10,662 -9,531 -4,573 608,026 -483,286 -27,955 -5.7
3 33,737 -32,104 -6,481 10,569 -9,817 -4,096 626,267 -475,162 -15,870 1.9
4 34,749 -32,104 -6,611 10,470 -10,111 -3,608 645,055 -466,622 -4,615 5.2
5 35,791 -32,104 -6,743 10,364 -10,414 -3,106 664,407 -457,645 5,862 6.8
10 41,492 -32,104 -7,445 9,725 -12,073 -405 770,230 -405,383 48,067 8.9
15 48,100 -32,104 -8,220 8,861 -13,996 2,642 892,907 -338,312 76,654 8.8
20 55,761 -32,104 -9,075 7,704 -16,225 6,061 1,035,12 -252,235 95,221 8.5
25 64,643 -32,104 -10,020 6,166 -18,809 9,876 1,188,438 -141,769 104,012 8.2
30 74,939 -32,104 -11,063 4,134 -21,805 14,101 1,350,896 0 105,710 7.9
forever 142,450 7.3
20
Table 3 IRRs for Various Growth Rates of Price gP and Rent and Expenses gR
IRR (%)
gP gR 10 years 20 years 30 years forever
1% 1% 2.70 3.97 4.06 3.66
2% 2% 6.07 6.43 6.12 5.63
3% 3% 8.89 8.52 7.86 7.26
4% 4% 11.34 10.28 9.45 8.78
1% 3% 3.91 5.78 6.22 7.26
0% 4% 1.45 5.40 6.80 8.78
21
Table 4 IRRs and NPVs for Stochastic Simulations with Fixed and Adjustable Mortgage Rates
Fixed Rate Adjustable Rate
Horizon median median VaR median median VaR
(years) IRR (%) NPV ($) NPV ($) IRR (%) NPV ($) NPV ($)
1 -33.06 -43,092 -117,177 -33.09 -43,166 -117,042
2 -7.66 -32,103 -130,718 -8.30 -33,503 -132,823
3 0.14 -21,896 -136,509 -1.44 -27,454 -144,466
4 3.66 -12,253 -138,645 1.43 -23,381 -153,885
5 5.49 -3,406 -138,482 2.89 -20,579 -161,281
10 8.03 32,464 -123,638 4.86 -16,259 -183,029
15 8.18 56,558 -103,177 5.24 -17,004 -190,343
20 8.02 72,581 -84,572 5.44 -17,404 -189,361
25 7.80 81,714 -69,551 5.57 -17,631 -183,603
30 7.57 84,670 -58,500 5.66 -17,048 -175,580
forever 7.23 139,202 28,857 6.23 35,422 -89,513
22
-200,000
-100,000
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
0 2 4 6 8 10 12 14 16 18 20
Figure 1 NPVs for Different Horizons, 3% annual price increase
NPV ($)
Required Return (%)
5 years10 years20 years
23
••••••• •••• •••••••••••••••• ••••••• •••••••••• •••••• ••• ••••••• •••••••••••••••• •• ••••••••• ••••••• •••• ••••• •••• ••••••••
••• •• ••••••• •• • ••
•••
•
•••
0
5
10
15
20
25
0 1 2 3 4 5 6 7 8 9 10
Rent/Price (%)
10-year IRR (%)
IRR = -6.6 + 3.0*(rent/price)
Figure 2 10-year After-Tax IRR vs. Rent/Price
24
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0.045
-100,000 0 100,000 200,000 300,000 400,000 500,000
frequency
NPV
Figure 3 Net Present Value (NPV) if Buy for Keeps with Fixed 30-Year Mortgage
25